openaienmodel: gpt-5-mini-2025-08-07
Prompting fundamentals
Key Points
- Be explicit about the task
- Include context and files
- Iterate and refine prompts
Summary
Practical guide to writing prompts that produce reliable, actionable outputs from ChatGPT. Focus on a clear task definition, supplying relevant context or data, and specifying the desired output format and constraints. Use iteration and small-step decomposition to improve results.
Key Points
- Outline the task: use an action verb, state the goal, the audience, and why it matters.
- Provide context and data: attach files, include examples, cite sources, and state constraints (time, scope, tools).
- Specify the ideal output: format (table, bullets, code), tone, length, and any structural requirements.
- Iterate and test: try incremental prompts (Okay → Better → Best), compare outputs, and refine instructions.
- Break complex requests into smaller steps or chained prompts for multi-stage workflows.
- Ask for options and set priorities (accuracy vs. creativity vs. speed) to guide the model.
Quick checklist for engineers
- Define action, audience, and success criteria.
- Attach or summarize key data and constraints.
- Request exact format and give an example if possible.
- Run several iterations, evaluate, and refine the prompt.